The MDR requires medical device manufacturers to conduct a planned, systematic literature search as part of the Clinical Evaluation and Post-Market Surveillance (PMS) and Post Market Clinical Follow-Up (PMCF). How do manufacturers conduct literature searches in an MDR-compliant manner? This article provides practical recommendations from database selection to selection criteria - including some insider tips.

A literature search according to Regulation (EU) 2017/745 (MDR) has two objectives: To collect data on the medical device under evaluation and to find information describing the State of the Art (SOTA) of the medical field under consideration. Thus, it contributes to updating the risk-benefit assessment and provides a basis for presenting the state of the art.

How does a literature search according to MDR succeed in practice? We have collected frequently asked questions and answered them from our point of view.

Clinical Evaluation, PMS, PMCF - is one literature search enough for all of them?

The MDR mentions the literature search as a tool several times:

• PMS (Annex III, 1.1),
• PMCF (Annex XIV, Part B, point 6.2) and
• Clinical Evaluation (Article 61, 3.).

Therefore, many manufacturers wonder how many literature searches they need to conduct. The MDR does not provide a direct answer. Theoretically, there is nothing against conducting a single literature search for multiple purposes. In this case, however, it is important that the quality of this search meets all regulatory requirements. Since the MEDDEV 2.7/1 rev. 4 applies as a guideline for Clinical Evaluation, its specifications must be considered.

Important to know: If manufacturers perform only one literature search, e.g., as part of the PMS or PMCF, the time interval between this literature search and the update of the Clinical Evaluation must not be too long. So far, there is no official indication for the permissible maximum interval. In our experience, the upper limit is six approximately months.

Checklist for practice: Search strategy according to MEDDEV 2.7/1 rev. 4

In terms of the practical implementation of an MDR-compliant literature search, this checklist provides support by answering common questions.

1. Databases

There is no consensus on the number and type of databases to search. However, according to the MEDDEV guidance document (see Appendix A4), it is not sufficient to search only MEDLINE or PubMed. At least one other database besides PubMed/MEDLINE should be used - for example:

• Embase,
• Cochrane Database of Systematic Reviews,
• ClinicalTrials.gov,
• Livivo,
• EBSCO,
• Cochrane CENTRAL trials register,
• WHO International Clinical Trials Registry Platform (ICTRP),
• Prospero international prospective register of systematic reviews.

In this context, a common question is whether Google Scholar is an acceptable database for a systematic search according to MEDDEV 2.7/1 rev. 4. Google Scholar does have the great advantage that it also searches for relevant hits in accessible full texts. However, disadvantages are the unpredictable repeatability and the impossible traceability of the search. It is acceptable to include Google Scholar as an additional database in the search. However, it should have been previously searched in at least two of the above databases.

Important to know: A total of at least two databases must be searched. The scope of the search depends on the product, its risk class, and the duration of its availability on the market. If you search multiple data sources, your chances of finding information about your products are higher.

2. Search terms

After you have decided on the databases, in the next step you should define suitable search terms. Their formulation is crucial for the quality and scope of the search. They should consider both the important aspects of the medical device’s clinical application and the specificity of the search:

• To find clinical data relevant to the product, it makes sense to include both the brand name and the manufacturer’s name in the search terms.
• When searching for State of the Art (SOTA) information, it is recommendable to use terms
• from the relevant medical field,
• for the intended purpose,
• for the indications, and
• that describe the product type, aspects of performance and safety of the medical device being evaluated, and marketing pledges.
• In addition, it is useful to search for brand and manufacturer names of similar and like products.

General search terms allow a comprehensive search, but often provide many results. In contrast, complicated and specific expressions greatly reduce the number of publications found and there is a risk that important data will remain undiscovered. A possible solution is constructed search terms using the logical operators "AND", "OR" and "NOT". They offer a certain flexibility and help to realize a balanced search.

Important to know: Search engines in PubMed and other scientific databases pre-modify the search terms before the search process. For example, if the term "product" is entered in PubMed, the underlying search algorithm uses related terms and searches for instead: ("economics"[MeSH Terms] OR "economics"[All Fields] OR "production"[All Fields] OR "productions"[All Fields] OR "efficiency"[MeSH Terms] OR "efficiency"[All Fields] OR "productivity"[All Fields] OR "product"[All Fields] OR "product s"[All Fields] OR "productive"[All Fields] OR "productively"[All Fields] OR "productivities"[All Fields] OR "products"[All Fields])

For a precise and reproducible search, the search queries should always be checked and the instructions on the respective database website for query creation should be considered.

3. Selection versus appraisal

Finally, it is about the selection and appraisal of the publications. What is the difference?

Selection
In the first step, the publications found are checked for relevance based on their abstract. For this purpose, (pre)defined selection criteria are used. They contain ...

• either exclusion criteria only: What is not excluded is included.
• or inclusion and exclusion criteria, which must not contradict each other.

Caution: Some Notified Bodies put a lot of emphasis on the logic behind the selection criteria.

Important to know: When sorting out publications, the respective reason for exclusion should always be documented. With this transparency, you fulfill the MDR requirement that both favorable and unfavorable data must be considered in the Clinical Evaluation (Annex XIV, Part A, point 2).

Appraisal
According to the MEDDEV, the selected, relevant publications must be evaluated based on (predefined) criteria (the "appraisal" criteria). The respective full text (to be acquired) of the selected publications is the basis of the appraisal. The publications considered in the Clinical Evaluation are assigned to different levels of evidence - for example, based on the type of publication, if applicable, the design of the clinical trial, the case number size, and the similarity of the medical device under consideration to the one being evaluated. This classification helps in evaluating the data or presenting the state of the art, especially in the case of conflicting information. Key arguments should be based on highly ranked publications.

Modern software solutions can help save time and increase efficiency in many of the steps described above. A variety of solutions are available for this purpose: From comprehensive databases with integrated statistical functions and intuitive interface design to the integration of artificial intelligence to support users in the selection and evaluation of publications.

Conclusion

As part of the Clinical Evaluation, PMS and PMCF, MDR requires manufacturers to conduct a systematic literature search. These searches can be bundled under certain conditions. A meaningful search strategy according to MEDDEV 2.7/1 rev. 4 requires a careful selection of databases and appropriate search terms. Finally, meaningful selection and appraisal criteria for the publications found must be defined and documented.

Are you currently facing the challenge of conducting a systematic literature search as part of the clinical evaluation of your new medical device? We will be happy to support you in your request. Even if you have questions about another complex topic in the field of clinical affairs, our team is there for you. Together with you, our experts will find the best answers to your individual question. Our initial consultations are non-binding and free of charge.

Best Regards